About

Hi! My name is Qurat ul ain. I recently (July 2026) completed my doctorate (PhD) in Computer Science at the School of Computer Science University of St Andrews, Scotland and I am looking forward to the next stopover of my professional journey!

My research focuses on data-centric AI and automated collaborative decision making. I am particularly how machine learning systems can reason about their own ignorance. In My PhD, I designed and implemented a novel evaluation framework for classifying model failures based on uncertainty (incomplete information), distortion (misrepresentation), and absence (lack of information), to reduce the cost of model training while maximising learning. The framework mapped theories of error from cognitive science to machine learning models. I showed that actively selecting training instances based on these error synergies outperforms conventional methods like query-by-bagging, and that the optimal error synergy depends on the dataset. The framework can be used as an interpretability tool to track the model’s learning in terms it its knowledge and errors.

My earlier research has focused on explainable automated decision making. I developed and implemented a collaborative interaction protocol for automated decision making between medical experts, encompassing human-human, AI-AI and human-AI collaborations. In my MSc. thesis at University of St Andrews, I investigated the techniques for justifying and exploring optimal treatment pathways for multimorbid patients as identified by SMT Solver Z3. I developed a general model of argumentation for justifying any path on a graph and a dialogue game between multiple agents that used the recommendation of the best path given by the solver. The dialogue generated explanations on the pros and cons of each node in the path as well as the reasons of not choosing alternative nodes in the path using argumentation. In my MSc. thesis at Maynooth University, Ireland, I developed a novel Virtual Research Environment for digital humanities researchers which provided a collaboration solution based on decentralized repositories using Dropbox as cloud storage. It also provided visualization and manipulation of relationship metadata between digital artefacts across these repositories. The user interface design employed \emph{Activity theory}, a social and psychological theory which relates individual consciousness to its social context through everyday interactions, formalised as \emph{activities}. It employed three-tier architecture and Model-View-Controller pattern. The client worked as a state-machine to interact with a RESTful server. The project also demonstrated an approach for verifying the client as a state machine using the SMT solver Z3.

Before my PhD, I spent 6+ years as a software engineer in industry developing CAD/CAE software at Bentley systems, Pakistan and front-end development at Software Competence Center Hagenberg (SCCH), Austria; and 2 years as a researcher at the Artificial Intelligence Research Institute (IIIA), Barcelona and Warsaw University of Technology (Poland). I have built experimental pipelines in R, Python/scikit-learn, JavaScript and production systems in C++/C#.

Reasearch Interests

• Data-centric AI for industrial agents and applications.
• Explainable and Error-aware AI
• Active machine learning
• Human-AI collaboration
• Multimodal information aggregation in multi-agent systems
• Automated reasoning

Grants and Fellowships

• University of St Andrews School of Computer Science PhD Scholarhsip: 08/2022 - 01/2026.
• Marie Skłodowska-Curie ITN-ETN Early Stage Researcher Grant: 09/2020 - 08/2022.
• Erasmus Mundus MSc. Scholarship Award: 09/2016 - 08/2018.

Academic Distinctions

• University of St Andrews’s Best Dependable Software Systems Project Award: 07/2018, DESEM Summer School, Ireland.
• First class honors in MSc.
• 6th position in a class of 44 in bachelor’s degree.